Re: [Dime] Problem with Origin- & Destination-Realm AVPs in RFC3588bis

Hi Fortune,

On Mar 5, 2009, at 4:46 AM, Fortune HUANG wrote:

[snip]

>
> My conclusion after comparing the grammars of the three RFCs:
> 1) According to the above RFC4282 grammar, "2.a " is a valid realm.

Correct.

>
> 2) According to the above RFC4566 grammar, "2.a " is not a valid  
> FQDN since
> it has only 3 characters (not 4 or more).

First, RFC4566 ABNF is not in a role for defining FQDN.. it is an ABNF  
for SDP grammar. So if the SDP grammar ABNF is wrong, it is not the  
problem of original FQDN ABNF. Besides, using "2.a" as an example is  
misleading. There is no root zones that are one character long (see  
ICP-1, RFC1591). The shortest root zone is two characters, which would  
e.g. be "2.ac" and this is correct according to the ABNF in RFC4566.  
The RFC1035 BNF would allow one character root zones, however, those  
just do not exist in Internet DNS.

>
> 3) According to the above RFC1035 grammar, "2.a" is not a valid  
> domain since
> it doesn't start with a letter (but a digit).

RFC1101 updates RFC1035 and relaxes the issue with a digit being the  
first character.

> If one could prove that the grammar of realm is the same as the  
> grammar of
> FQDN,  then, RFC4282, RFC1035 and RFC4566 would be proven inconsistent
> according.

So far, no problems with cases 2) and 3). Regarding the case 1) few  
notes. RFC3588bis section 1.3. states that "NAI realm names are  
required to be unique, and are piggybacked on the administration of  
the DNS namespace." This basically means one loses its rights for  
"creative" realm names when used with Diameter. In DNS, one character  
root zones do not exist, thus "2.a" is not legal within Diameter scope.

> However, I am not sure if I have found the right place where the  
> strict
> grammar of FQDN is defined. Please tell me if you know.
> But RFC4566 and RFC1035 were the materials my comment in the  
> previous email
> was based on.

Although this stuff is spread a bit around and topped with de-facto  
assumptions, I think there is no issue.

Cheers,
	Jouni

>
>
>
> Best Regards,
> Fortune
>
> -----Original Message-----
> From: Victor Fajardo [mailto:vfajardo@tari.toshiba.com]
> Sent: Thursday, March 05, 2009 6:02 AM
> To: lionel.morand@orange-ftgroup.com
> Cc: fqhuang@huawei.com; glenzorn@comcast.net; dime@ietf.org
> Subject: Re: [Dime] Problem with Origin- & Destination-Realm AVPs in
> RFC3588bis
>
> Hi Fortune,
>> I'm not sure to understand but I might have missed something.
>>> From a syntax point of view, what is the difference between a FQDN  
>>> and a
> realm?
>> What would be the "potential" impacts to say that the  
>> DiameterIdentity can
> be a FQDN or a realm?
>>
> I have the same question as Lionel. Syntactically, FQDN and realm  
> are the
> same from the parsers point of view. The difference is in semantics  
> which is
> already specified by the AVP having that type.
>
> regards,
> victor
>
> _______________________________________________
> DiME mailing list
> DiME@ietf.org
> https://www.ietf.org/mailman/listinfo/dime